This paper introduces a mechanism for deployed quickly in the event of a disaster and quickly exchange the battery of a drone. This mechanism does not require a platform like the conventional battery exchange mechanism and exchange the battery with the force as the drone descends. Quickly exchange the battery is possible by simultaneously attaching and detaching a charged battery and an empty battery. A prototype will be manufactured, checked operation and a power experiment. A new battery exchange method for the drone will be performed.

In unmanned construction, the landmark that the drone shoots is ambiguous. In order to determine this, it is necessary to grasp the cognitive space during the work of the operator. Therefore, we investigated and analyzed the landmark recognition characteristics during the work, using the fusion method of SAGAT (Situation awareness global assessment technique) and direct drawing method.

We are developing machining robot applying new parallel manipulator system with our folding convex type telescopic manipulators. Compared with the conventional linier actuators, our convex type manipulator can make long reach and are very light. They are also compact when they are reeled up. However, our conventional manipulators are insufficient on the power. In addition, there are some problems in the structure, and it cannot be used stably. Therefore, in this study, we redesigned the conventional telescopic mechanism, manufactured a telescopic manipulator type-K that can be used easily with enough high output for parallel manipulator, and measured and evaluated its performance.

Rescue robots for searching people under the rubble of collapse buildings is still in high demand, by many countries that suffer from natural disasters such as earthquakes. An articulated robot configuration is advantageous for such purpose. In this research, we tried to combine the advantages of an articulated robot and Archimedes screw for moving inside rubble and over rough terrain. In this paper, we introduce the basic concept and show some experiments results of the developed robot.

Chatbots that respond according to keywords have been proposed. However, if the keyword is not registered, there is a problem that proper conversation cannot be performed. In this paper, we propose a method to use a registration response sentence using word2vec.

This paper presents a communication robot system with a simple LED display for representing timing for turn-taking in human-robot interaction. Human-like conversation with non-verbal information, such as gestures, facial expressions, tone of voice, and eye contact, makes a contribution to provide more natural communication. If the robots could use such verbal and non-verbal communication skills, it would lead to establish a social relation between a robot and human. Timing for turn-taking in human communication can be important non-verbal cues to efficiently convey messages and to share opinions with each other. In this study, we show some experimental results to discuss the effect of response timing for turn-taking on communication between a person and robot.

In recent years, the use of 3DCG and 3D avatar has increased with the development of image processing technology in various digital contents. Motions in 3D avatars has become very complicated, and it takes a very long time to create them since they are made manually in most cases. In this study, estimating emotions from sentences for the purpose of generating 3D avatar motions automatically from the estimated emotion information and the semantic information of the sentences. Thereby, it can save a lot of time. As the first stage of the research, estimation of emotion from sentences such as dialog texts is investigated.

We, humans, have an instinct of a social nurturing of children. We are researching a robot system that can gather and process necessary information effectively in the real world with the help of people. By using an android with a child-like appearance, the system might keep eliciting active involvement from the human side, just like an adult takes care of a child instinctively. This study focuses on walking hand in hand with the child as an example of the natural behaviors of child and adult. By walking hand in hand with humans, the android can convey intention, e.g., direction to go, by the physical interaction without having a conversation. In the experiment, the subject is asked to follow the android with hand in hand or side by side. As a result, we found a difference in the ratio of gazing to the android in each condition.

Guidance is an important task for robots that provide services. In this paper, we focus on a robot with a projection function, and propose a method for naturally guiding people without explicit instructions by combining the movement of the robot and the enlargement / reduction of the projection size. The experimental results showed that the subject's movement was affected by both the movement and the projection of the robot, confirming the effectiveness of the proposed guiding action.

In this paper, we propose the mechanism of a 4-dof panda-like expression robot capable of interacting with human beings. To express emotions on human-robot interaction, the robot has pressure sensors on the body and the head, a camera on the body, a microphone on the nose, and a speaker on the body. Also, a fuzzy-based emotion recognition algorithm for three emotions such as happiness, sad and anger is discussed. Using the robot, emotion expression experiments are conducted, and the effectiveness of the robot mechanism and the emotion expression system is confirmed.

It is important to consider non-verbal information to understand the situation in speech. The purpose of this research is to develop robot partners for providing recommendations based on non-verbal communication. In this paper, we propose a system that recognizes surrounding objects and humans’ expressions for recommending appropriate topics to users. The proposed system consists of two convolutional neural networks (CNNs) for performing object recognition and human expression detection respectively. The facial expressions and objects present in the surroundings are used as non-verbal information, and the provided topics are hierarchical. We first evaluate the proposed system through objects and human expressions recognition accuracy. With this two non-verbal information, the proposed system able to provide relevant recommendations to users according to their situation.

Eating with others enriches our dining experience and has many social functions. However, more and more people tend to eat alone these days because it is difficult to match meal times and places with others. In this paper, we developed the mixed reality system for co-eating with a robot. In this system, a robot and a mixed reality headset are connected and people who wear the headset can see that real robot puts computer graphics food into its mouth as if eating. In the evaluation of the developed system, people over 13 years old can enjoy eating more, feel more delicious and eat more when a robot eats together than when it only talks and does not eat.

Establishing a smooth communication between robot and human is an important issue for communication robots. In recent years, several studies have adopted robot’s emotion expression to realize such communication. In previous research, emotion of the robot was expressed by several factors such as robot’s language, voice, facial expression, and body motion. In this paper, the vertical oscillation of the robot was proposed as an effective factor to express robot’s short-term emotion during a conversation. For this purpose, we prepared a robot expressing its basic four emotions, i.e. happiness, anger, sadness, and relief, utilizing a gesture and a vertical oscillation. To verify the effect of the oscillation, we conducted an experiment comparing the subjective evaluation of human about robot’s emotion expression with/without oscillation. The result of the experiment showed that each emotion expression has proper oscillation.

Robot’s communication function is important for coexistence of robot with human. Recently, several research have been conducted to facilitate the communication between robot and human. In this paper, the visualization of robot’s utterance information was proposed as a way to convey robot’s intention more clearly as well as increasing human’s subjective evaluation about the robot, utilizing a dialogue projection system. In this system, after the talk of the robot, some words from robot’s talk was projected on the wall behind the robot by a projector. In order to verify the performance of the proposed system, a comparative experiment was conducted under three conditions: projecting, subtitling and not showing the information of the robot’s talk. The result of the experiment showed that subtitling the talk information improves the subjective evaluation of human about the robot.

Recently, AI equipped robots, smart speakers have appeared in homes, opportunities to come into contact with robots in general households and towns have increased, and various people also have a high possibility of controlling robots. To be able to control the robot smoothly, the robot side needs to understand human intention. In this paper, we propose a system that controls a robot from human ambiguous instructions, images of feet and object detection. In this system, Jetson Nano, USB microphone and USB camera are attached to the mobile robot for control. Control of the robot is performed only with body direction estimated by feet image of the user, voice instructions and Environment recognition using object detection without attaching special sensors. Ambiguity in speech recognition that do not limit the target using the directive are recognized.

In this report, we measured the distance between the human and the robot and the loudness of the movement sound during the dialogue in order to quantitatively analyze the effect of the robot's activity on the human's concentration. In the experiments, it was assumed that humans were basically stationary and engaged in dialogue, and the distance was measured while subjects perform two types of tasks that compose the dialogue. The significance of the result was verified by t-test. As a result, it was confirmed that the main factors that hinder human concentration due to the movement of the robot are due to hearing, and that the dialogue does not depend on the state of the dialogue, such as speaking or listening. From the measurement results, the approximate function used to calculate the size of the formation space was calculated.

There is a growing expectation and use of robotic devices, especially communication robots, in the nursing care of the elderly. One of the features of the communication robot is that they do not just support the movements of elderlies like other nursing care robots but also support their social activities. For these reasons, the application, function and efficiency of the communication robot to the elderly are difficult to understand or even evaluable of its validity. Therefore, we propose a new method based on ICF to classify applications, functions and efficiency by capturing features that communication robots have.

This paper describes sonic communication using an electrostatic driver. It can send and receive audible and inaudible sounds. This device is intended to communicate between humans and robots, or between robots, using sound waves. To verify this device, we transmitted and received sound waves using two identical devices. The frequencies are 1kHz and 25kHz. as a result of this experiment, it was confirmed that sound wave communication in the audible range and non-audible range was possible between devices.

Atmosphere estimation system for family-type robot in living space is proposed to realize smart symbiotic space, where a three-dimensional atmosphere (atomosfield) is calculated based on multiple human emotions recognized from webcam video. The system is implemented with OpenCV and Dlib for face and emotion recognition, and tested for webcam arrangement. As a result of the experiment for thee human and two LOVOTs interaction in the living space, the proposed system has shown the highest detection of times of 380 at the level of 0.45[m] from the floor and the front (0 deg) of the users in face recognition. The proposed atmosphere estimation realizes casual and delicate support service for people in smart living environment.

Toward the achievement of natural attentive listening using an android, a nodding synchrony behavior with a human talker is an effective nonverbal behavior from the finding of several works. In addition, according to the related work, a human listener’s eyeblinks synchronizes with an android talker’s one in face-to-face communication. Here, we hypothesize if an android which is in the role of a listener imitates eyeblinks and nodding simultaneously a human’s one in face-to-face communication, the android can act more attentive listening behavior. In this paper, we propose a method to imitate human eyeblinks and nodding using an android to synchronize the human’s one. Through the experiment, we confirmed that the android can generate the imitation behaviors of human eyeblinks and nodding.

The purpose of this study is to propose an intervention system that can continuously measure grip strength using a communication robot. And we verified the effect of this system. As a result, it was confirmed that the measurement behavior for three weeks could be continued by the face recognition and the voice call functions installed in the communication robot. It was suggested that grip strength is an effective index for health management of the older adults living at home. This study has shown that the communication robots are an effective tool for encouraging the continued measurement of healthcare monitoring systems.

This report discusses an objective evaluation method for tool-incorporation in the field of assistive products. We focus on a bypass prosthetic hand, which enable a sound hand person to experience a prosthetic hand. We expect that the result of this work to contribute to the further discussion on the training method of the prosthetic hand.

The number of physically handicapped people in Japan is about 1.93 million, and most of the people with walking disorders use wheelchairs on daily basis. Living in a wheelchair on daily basis reduces muscle strength, causing disuse syndromes such as muscle atrophy, joint contracture, and decreased mental activity. In order to prevent these syndromes, standing training and range of motion training are effective methods, but they are a heavy burden on caregivers. In addition, there is a survey result that excretion assistance is the most burdensome thing in nursing care. Therefore, in this study, we developed toilet activities support wheelchair which has three functions of standing support, passive range of motion exercise of the hip joint, and toilet activities support.

To contribute to the prevention and rehabilitation of knee joint injuries, we previously developed an apparatus called ERIK (Electro-Rheological-fluid-operated Iso-Kinetic exercise equipment for the lower limb) using electrorheological (ER) fluid brakes which can control the resistance torque rapidly and reversibly. We also designed a load that allows the swinging leg to move at a constant target velocity while providing a constant resistive force of over 200 N to the sliding leg in another control mode. Other types of training require equipment that exercises both legs with individual control of each leg. We are thus developing equipment (called EBMO) for bilateral movement training with a horizontal resistance force provided by ER fluid.

In Japan, aging of the population is happening. Due to this, shortage of caregiver is one of the problems of aging and it is said that 380,000 caregivers are required in 2025. As one of their tasks, caregivers must check diapers every 2 hours whenever it is day or night. This is the biggest burden for care givers, especially at midnight. It is also a burden of care receivers because it makes their sleeping quality poor. Usage of excretion detector devices may solve the problem. However, there are inconveniences in collecting and sanitizing the device for re-using and discomforts of attaching large device to a diaper. To avoid caregivers and care receivers from a burden, two functions are needed: disposability and ultra-thinness. This research developed and evaluated disposable film-based urine detector device.

General classification methods cannot treat unexpected patterns that are not considered in the training process. For this problem, our research group has proposed a probabilistic classification method that can classify abnormal patterns as the unlearned class. For the implementation of the classifier in the embedded hardware, this paper proposes a novel approximate Bayesian classification method with the anomaly detection based on Gaussian mixture models and the probabilistic density function of the unlearned class. The proposed classifier was applied to forearm motion classification in the experiments. Experimental results demonstrate the proposed method can achieve high classification performance as same as the previous model, and the effectiveness of the proposed method could be confirmed.

Many people with physical disabilities are forced to live in wheelchairs, and it is extremely difficult to perform daily living activities (ADL) without assistance. Therefore, it has been proposed that the ADL can be performed without assistance by mounting a welfare robot arm on a wheelchair. It is hoped that it will be possible to support independence of the physically handicapped. In this research, it is thought that various tasks can be executed by expanding the operation area of the robot arm. For tasks such as picking up and carrying objects as well as picking up objects that have fallen to the floor, the pitch of the mounting section has been rotated to enable intuitive operation from the operator's standpoint.

We have developed a sensor that contains three photo-reflectors and can measure the size and direction of pressure of silicone rubber film, and have developed a controller that generates information by the tongue. At first, this was used as a computer mouse to verify its operability. After that, the possibility of application as a communication device to a machine was examined. The first is a system in which a robot arm is used to hold a substrate for a healthy person, and a robot is used to assist soldering work using both hands of a human, and the operation is performed using a tongue. The second is a system for steering electric wheelchairs for the disabled. We evaluated the usefulness of the three-element oral soft input device through these two studies.

Sit-to-stand motion (hereinafter, this is called “STS”), which is one of daily motions, is a heavy burden on elderly people. In this laboratory, a movable handrail type STS measuring equipment has been developed and prototyped. We analyze the STS of the user from the measured values of the sensors installed in this measuring equipment. In this paper, in order to consider the improvement of the support method by the STS motion measurement device, the measurement value of each sensor by the handrail drive method which was set focusing on the speed of the standing-up motion and the positional relationship between the handrail and the body was examined and evaluated.

This study proposes a novel stair-climbing wheelchair using lever propulsion control by human upper body. The developed wheelchair is designed based on a regular manual wheelchair, which has manual wheels with casters for planar locomotion. It also has a rotary-legs mechanism operated by a user through a lever. In addition, the wheelchair has a posture transition mechanism to shift a whole center of mass between desired positions for planar locomotion and stair-climbing. In this paper, we investigate muscle loads of a user by different lever propelled directions during stair-climbing. Furthermore, we present stair-climbing experiments with a person with spinal cord injury to verify the effectiveness of the proposed stair-climbing methodology.

Regarding the wheelchair climbing over a step, a wheelchair user cannot get over the step without the help of an assistant. Therefore, in order to reduce the burden on the caregiver and extend the range of action of the wheelchair user, we propose a step-crossing assist mechanism that allows only the wheelchair user to cross the step. A crutch-type assist mechanism has been developed to overcome the front wheel of a wheelchair, and a double-boost assist mechanism has been developed to overcome the rear wheel of a wheelchair. The crutch-type assist mechanism is configured to push over a step to form a support rod to get over a front wheel, and the double-boost assist mechanism is to extend a link by pushing down a handle portion of the assist mechanism and push up a rear wheel. The effectiveness of the assist mechanism developed by experiments was confirmed.

To assist stair climbing movement for ALS patients and patients with spinal cord injury, we are developing a stair climbing machine. Although most stair climbing machine adopt crawler mechanism, crawler system has some problems such as damaging floor and slip dropping. Therefore, we adopt linkage mechanism for stair climbing. This device has advantage that users can keep their posture horizontally．

In this research, a concept of robotic wheelchair which automatically follows a human caregiver side by side is proposed. This concept is aiming to promote the communication between a wheelchair rider and his/her caregiver since they can talk face-to-face while advancing forward. This research focuses on a derivation of a motion control law to achieve the autonomous side-following by the robotic wheelchair, determination of parameter values of virtual spring and dampers connecting the robot and caregiver to establish the side-following which is suitable for comfortable communication, based on an impedance method. The determination of the parameter values has been done by considering the two physical conditions regarding the communication, loudness of the voice, elevation angle of one’s neck to see other’s face, and convergence of the relative position and orientation are considered. Finally, an experimental prototype robotic wheelchair having a LRF and a bird’s eye view camera, and four-active caster modules is fabricated to perform side-following motion control experiment in a room-environment. The obtained results demonstrated that the prototype wheelchair robot converged to a commanded relative position against a human subject, and proposed the feasibility and effectiveness of the proposed method.

In this study, we investigated the development of a wheelchair leave-warning device for patients with high needs from the care site. That is, it detects the seated state of the wheelchair user, and issues a warning as necessary. In this report, we show the system implementation to the wheelchair.

Low back pain is a serious problem in Japan, so analysis of lumber load is needed. The conventional analysis method is an invasive analysis, and a non-invasive and safe analysis method is required. The purpose of this study is to make a physical model of L4-L5 vertebral body from radiographs and to estimate the lumbar load from the model. Five X-ray images of the upper body angle, standing postures, and backward bending were taken for each subject, and a three-dimensional model of the L4-L5 vertebral body was created. FEM analysis was performed using the model created, and the load acting on the intervertebral disc during the photographed postures ware measured. The results were evaluated for each of the four subjects to measure individual differences. Based on these results, we approached the lumbar load analysis using a physical model, which is the goal of this study.

In this research, we examined an operation support method by gaze input using gaze amount histogram. In order to accurately identify the direction the user wishes to move and the direction which the attention is directed, in the gaze driving, estimation system based on the gaze amount histogram was constructed and a running experiment was performed. As a result, showed no significant difference in the traveling route compared to the conventional method, but the subjects’ questionnaire resulted in a higher score than the conventional method in terms of safety. Therefore, it was possible to confirm the effectiveness of the operation support method using the gaze amount histogram.

Hand-Eye-Coordination (HEC) is one of the most important abilities in sports. It is the ability to quickly and accurately output to movement from visual information. And it is deeply involved in badminton smash receive. Therefore, it is an indispensable ability for badminton players, and it is necessary to evaluate HEC appropriately. However, in previous studies on HEC evaluation, HEC is evaluated from final result. As a result, information processing is a black box. So, it is difficult to determine which phase of perception or movement has a problem in HEC.

Therefore, the purpose of this study is to identify the factors of perception phase in badminton smash receive.

In this research, we aim to elucidate “Kire (Sharrpness)” in Karate movements using wearable devices. If we could understand important points in karate movement, it is considered for us useful for guidance and evaluation about Karare practice and competition. Previous report, we verified the “Seiken-duki” (Fore fist punch), which is the basic movement of Karatedo. We uses three devices which can measure acceleration and angular velocity[1]. However, the large measuring devices put a burden on performing the Karate operation.

In this paper, the aim is to make the devices smaller. We verify whether the same result as the first report can be obtained by using the create small devices. In the experiment of this paper, we also verified the motion of hand’s wrists and waist movement for an experienced player and a first time player.

In order to examine the training method of soccer feint motion, the relationship between the visual effect and the prediction ability of the defender has been verified in this study. The experiments for acquiring the gaze behavior and the prediction result using the temporal occlusion method which is a prediction ability evaluation method have been conducted. As the results, it was cleared that the significant visual effects are included in the early frames of feint motion. In addition, skilled sportsmen who had the ability to quickly and accurately prediction, to place fulcrums around the feet. These results suggest that there are effective predictive gaze strategies for feint motion.

In the previous study, an experimental system to measure propulsive force during swimming using the humanoid robot SWUMANOID was constructed. However, the recovery motion of crawl stroke was not appropriate because of the earlier timing of the rolling motion, which was caused by an attaching device to SWUMANOID. Also in the previous study, the better stroke which can improve the rolling performance in those experimental condition was found by an optimizing calculation. The objective of this study was to examine the validity of the stroke obtained in the previous study by a free swimming experiment. Two types of improved stroke were created and free swimming tests were conducted for those strokes. Since the significant improvement in the rolling performance could be found for the improved strokes, the validity of the proposed strokes were confirmed.

A lot of research has been done on striking sports, many of which are burdensome to players because players have to be put markers on such as optical markers and accelerometers. Therefore, it is necessary to obtain useful data for feedback to players only using multi cameras. In this paper, we propose a method for motion capture and calculating the impact force using synchronized high-speed cameras and video cameras in golf. We performed the experiment with one subject, and it was found that it was possible to calculate the joint torque from the impact force to the wrists and evaluate the posture at the impact.

Bicycle is a tool in everyday life that allows you to travel farther and faster.It is important to consider a design that allows the human musculoskeletal structure to efficiently transmit the forces that can be generated to the bicycle.The purpose of this research is to constructively elucidate how humans generate force when pedaling by using a robot that imitates the human musculoskeletal structure. For the analysis, in this report, we developed a robot driven by pneumatic artificial muscle and realized a pedaling by using a pressure signal inside the artificial muscle as a trigger.As a result, pedaling was realized by using the pressure drop as a trigger.

Pedaling skills are one of the important factors to bring out one’s best performance in pedaling exercise, and refer to the ability to efficiently convert bicycle pedal effort into a driving force. It has been reported that pedaling skills are strongly connected with the muscle synergy of lower limbs. However, a quantitative evaluation method for this connection has not been established yet. The objective of this study is to establish a quantitative evaluation method for pedaling skills. In this study, we proposed a method to derive muscle synergy by measuring and processing the signals of surface electromyography of the lower limb during pedaling exercise, where both the temporal structure and the space structure of the muscle synergy can be visualized. In this paper, we apply a machine learning method to the visualized muscle synergy data in order to estimate the pedaling skill level of a test subject.

This paper proposes an image analysis method of recorded images of long jump captured with monocular camera. The posture at the moment of jumping is automatically recognized based on the running lanes and the spring board, and the various features for analysis at that moment are extracted and analyzed. The proposed method is applied to the actual data of experienced players and some features are found to be correlated to the jumping records.

In the field of competitive cycling, the importance of improving own pedaling skills has been recognized. In addition, highly trained cyclists argue that the appropriate motion of their upper limbs is needed to make the pedaling exercise more effect. Thus far, few studies have investigated the roles of the upper limb during pedaling exercise. Then, this study aims to experimentally investigate the roles of upper limb by measuring the handling motion. This paper describes the experimental environment, the signal processing method of sEMG to derive the muscle synergy, and the experimental results.

In this paper, we measured and analyzed bowling forms using the video analysis software Kinovea, in order to identify similarities between senior bowlers. From measurement results, standard deviations of time spent on throwing in senior subjects were smaller than those in beginner subjects. From measurement results of foot positions, similar trajectories of foot positions were observed in senior subjects.

The purpose of this study is to develop a video-based motion analysis method using OpenPose for evaluating the performance of sport climber. To evaluate the skills of sport climber quantitatively, we proposed three kinds of evaluation indices concerning the smoothness and stability of motion as well as the effort of arm. The result of experiment shows that these indices are effective to distinguish the level of climbing. In addition, we developed the testing device of an eccentric grasping-force to measure the strength of the climber’s grasping. The measurement result shows that the characteristics of the eccentric grasping force clearly differ by the level of sport climbers.

To find out risk factors that explain risks of the anterior cruciate ligament (ACL) injury during sports activities still require more explorations. In this paper, we propose to use the auto-regression model (AR model) to fit their muscle force time-series data so that we can analyze athletes’ muscle cooperations. We measured their physical motions during Drop Vertical Jump (DVJ) test and acquired muscle force data from heel contact timing to toe-off timing for the second jump during the DVJ test and fit the AR model. We hypothesized that all muscles are activated soon after their contact timing and only reflections affect their muscle activities. Results revealed that not only reflections of the muscles but also voluntary controls of the muscles need to be considered.

Pneumatic drive systems are widely used in assistance devices and flexible drive robots owing to their characteristics of light weight, high output, and excellent flexibility. Compressed air generated by engine compressors and large motors is mainly used to drive these devices, but such large and heavy air pressure sources are barriers to the mobilization of pneumatic drive devices. Then ,we focus on the physical properties of flammable dimethyl ether (DME), and consider obtaining energy in two stages via a phase change and the combustion of DME to develop a new mobile air pressure source that can release enough pressure and flow. In this paper, we measured the pressure when the mixed gas is burned in a semi-hermetic state.

Hydraulic drive interests many researchers as a drive system that has high impact resistance. The hydraulic system can be roughly divided into two parts: an actuator unit that generates movement and a pump unit that supplies hydraulic oil to each component. The robot needs to operate within the pump unit capacity. However, it is difficult to predict the required pressure and flow rate during the operation of the robot. Especially in the case of legged robots, it is rarely performed because the state of the robot changes drastically in periodic operation. In this study, we estimated the required pressure and flow rate in the walking motion and designed the leg trajectory. The validity was shown experimentally.